Recent Shallowing of the Thaw Depth at Crater Lake, Deception Island, Antarctica (2006–2014)

Catena 149 (2017) 519–528

Contents lists available at ScienceDirect

Catena

journal homepage: www.elsevier.com/locate/catena

Recent shallowing of the thaw depth at Crater Lake, Deception Island, Antarctica (2006–2014)

M. Ramos a,⁎,G.Vieirab,M.A.dePabloc,A.Molinac,A.Abramovd,G.Goyanese a Department of Physics and Mathematics, University of Alcalá, Madrid, Spain b Centre for Geographical Studies – IGOT, Universidade de Lisboa, Lisbon, Portugal c Department of Geology, Geography and Environment, University of Alcalá, Madrid, Spain d Russian Academy of Sciences, Institute of Geography, Russia e Instituto de Estudios Andinos “Don Pablo Groeber” (UBA-CONICET), Universidad de Buenos Aires, Argentina article info abstract

Article history: The Western Antarctic Peninsula region is one of the hot spots of climate change and one of the most ecologically Received 19 January 2016 sensitive regions of Antarctica, where permafrost is near its climatic limits. The research was conducted in Decep- Received in revised form 11 July 2016 tion Island, an active stratovolcano in the South Shetlands archipelago off the northern tip of the Antarctic Pen- Accepted 12 July 2016 insula. The climate is polar oceanic, with high precipitation and mean annual air temperatures (MAAT) close to Available online 18 July 2016 −3 °C. The soils are composed by ashes and pyroclasts with high porosity and high water content, with ice- rich permafrost at −0.8 °C at the depth of zero annual amplitude, with an active layer of about 30 cm. Results Keywords: Permafrost from thaw depth, ground temperature and snow cover monitoring at the Crater Lake CALM-S site over the period Decreasing thaw depth 2006 to 2014 are analyzed. Thaw depth (TD) was measured by mechanical probing once per year in the end of Active layer January or early February in a 100 × 100 m with a 10 m spacing grid. The results show a trend for decreasing Snow insulation thaw depth from ci. 36 cm in 2006 to 23 cm in 2014, while MAAT, as well as ground temperatures at the base Antarctic peninsula of the active layer, remained stable. However, the duration of the snow cover at the CALM-S site, measured through the Snow Pack Factor (SF) showed an increase from 2006 to 2014, especially with longer lasting snow cover in the spring and early summer. The negative correlation between SF and the thaw depth supports the sig- nificance of the influence of the increasing snow cover in thaw depth, even with no trend in the MAAT. The lack of observed ground cooling in the base of the active layer is probably linked to the high ice/water content at the transient layer. The pyroclastic soils of Deception Island, with high porosity, are key to the shallow active layer depths, when compared to other sites in the Western Antarctic Peninsula (WAP). These findings support the lack of linearity between atmospheric warming and permafrost warming and induce an extra complexity to the understanding of the effects of climate change in the ice-free areas of the WAP, especially in scenarios with increased precipitation as snow fall. © 2016 Elsevier B.V. All rights reserved.

1. Introduction (Smith and Brown, 2009., Bojinski et al., 2014)andarekeyindicatorsof climate change in the Polar Regions (Anisimov et al., 1997; Burgess et Deception Island is located in the South Shetlands archipelago, north al., 2000), reﬂecting changes in the ground surface energy balance of the Antarctic Peninsula (AP), one of Earth's regions with strongest at- (World Meteorological Organization, 1997). The distribution and prop- mospheric warming signal, with an increase of ci. 3 °C in the MAAT from erties of permafrost and the ALT in the Antarctic Peninsula region are the 1950s (Marshall et al., 2002; Meredith and King, 2005; Turner et al., poorly understood (Bockheim, 1995; Bockheim et al., 2013). The situa- 2005; Turner et al., 2009; Turner et al., 2013). The environmental conse- tion started to change with the International Polar Year (IPY) 2007– quences of this warming on glaciers and ice-shelves have been widely 2008, which was the framework for an international effort to improve studied, but the consequences for permafrost, are only now beginning monitoring of permafrost and the ALT across the Antarctic (Vieira et to be investigated (Vieira et al., 2010; Bockheim et al., 2013). Permafrost al., 2010). temperature and active layer thickness (ALT) are Essential Climate Var- The Circumpolar Active Layer Monitoring (CALM) program is a net- iables as deﬁned by the Global Climate Observing System (GCOS/WMO) work of permafrost observatories distributed over both Polar Regions and selected mid-latitude mountain ranges (Brown et al., 2000). CALM ⁎ Corresponding author. is an international global-change monitoring program concerned with E-mail address: [email protected] (M. Ramos). ALT dynamics and the shallow permafrost environment, with CALM

http://dx.doi.org/10.1016/j.catena.2016.07.019 0341-8162/© 2016 Elsevier B.V. All rights reserved. 520 M. Ramos et al. / Catena 149 (2017) 519–528 sites in the Southern Hemisphere designated as CALM-S (Guglielmin, This paper aims at characterizing the active layer thermal regime 2006; Nelson and Shiklomanov, 2010). CALM-S sites are also integrated and climate controls, through the analysis of thaw depth (TD), air and in the Global Terrestrial Network for Permafrost (GTN-P, WMO/GCOS) ground temperatures and snow cover observations at the Crater Lake (Burgess et al., 2000). CALM-S site in Deception Island using data from 2006 to 2014. Low altitude permafrost temperatures are slightly below 0 °C in the South Shetlands, showing that the region is near its climatic boundary, 2. Study area thus being probably one of the regions with highest sensitivity to climate change in the Antarctic (Ramos et al., 2007; Ramos and Vieira, Deception Island (62° 55′ S, 60° 37′ W) is located in the South Shet- 2009; Bockheim et al., 2013). This fact reinforces the importance to lands archipelago in the Bransfield Strait, about 100 km north of the study the evolution of permafrost and active layer in the region. Antarctic Peninsula (AP) (Fig. 1). The horseshoe shaped island is a stra- The CALM-S protocol as an approach to standardize ALT monitoring tovolcano with a diameter of 15 km, a maximum elevation of 539 m asl has been implemented by our team in Deception Island, one of the at Mount Pond in the eastern part of the volcano rim and with a 7 km scarce areas of the South Shetlands allowing using mechanical probing wide caldera open to the sea. About 57% of the island is covered by gla- to measure thaw depth. Despite being an active volcano, the high inter- ciers and an area of about 47 km2 is glacier free (Smellie and López- stitial ice-content of permafrost and high porosity volcanoclastic soils, Martínez., 2002). favour thermal insulation and permafrost is ubiquitous almost down The climate of the South Shetlands is cold-oceanic with frequent to sea-level. Geothermal anomalies show only very local effects, mainly summer rainfalls and a moderate annual temperature range. MAAT along faults and within buffers a few hundred meters wide (Goyanes et are close to −3 °C at sea level and average relative humidity is very al., 2014). Therefore, the ALT conditions in most of the island are not af- high, ranging from 80 to 90%. The weather conditions are dominated fected by the geothermal heat flux. by the influence of the polar frontal systems, and atmospheric circula- CALM-S monitoring sites target at identifying the spatial variability tion is variable, with the possibility of winter rainfall (Styszynska., of the ALT and the controlling effects of parameters such as topography, 2004). The annual meteorological data describe two seasons corre- soil physics properties, vegetation, hydrology and snow cover. Observa- sponding to the annual cycle of soil freezing and thawing (Ramos et tions are usually carried out in a 100 × 100 m grid with sampling nodes al., 2012). Electrical resistivity surveying suggests permafrost thickness at 10 m spacing, where TD is measured by mechanical probing close to to be between 3 and 25 m in Deception Island (Vieira et al. 2008b; the end of the thaw season. In complex topography sites, smaller grids Bockheim et al., 2013; Goyanes et al., 2014). may be implemented. Arrays of shallow boreholes for monitoring As a result of recent eruptions, Deception Island is covered by volca- ground temperatures may also be used in areas where probing is not nic ash and pyroclasts, and many of the glaciers remain ash-covered possible and a meteorological station is used to analyze the climate con- today. Pyroclastic deposits covered the snow mantle, and buried snow ditions. At some Antarctic sites, this approach is complemented with is still present at some sites. Deposits are very porous and insulating deeper boreholes that are monitored for permafrost temperatures with high water/ice content near the surface, and give rise to a thin (Vieira et al., 2010). ALT, varying from 10 to 96 cm depth (Bockheim et al., 2013). On

Fig. 1. Location of Deception Island in the South Shetlands Archipelago (Antarctica). M. Ramos et al. / Catena 149 (2017) 519–528 521 lower valley slopes, exposures of fossil snow (buried ice) with perenni- ally frozen ice-cemented volcanic debris on top can be observed, testify- ing post-eruption aggradation of permafrost. At these sites, ice- cemented permafrost also occurs under the buried-ice layer. Buried- ice is widespread on Deception Island, especially along the lower slopes, and ice-cemented permafrost occurs almost down to sea level as shown by geophysical surveying and trenches (Vieira et al., 2008b). The Crater Lake CALM-S site is located in a small and relatively flat plateau-like step covered by volcanic and pyroclastic sediments at 85 m a.s.l north of Crater Lake (62°59′06.7″ S, 60°40′44.8″ W). The site was selected due to its flat characteristics, absent summer snow cover, distance to known geothermal anomalies, good exposure to the regional climate conditions (mitigating site specific effects and being representa- tive in a regional context) and also because of its proximity to the Span- ish station “Gabriel de Castilla”. The ground surface is completely devoid Fig. 3. General view of the Crater Lake CALM-S site, the arrows point the position of the of vegetation and the MAAT at the Crater Lake CALM-S site between 28/ camera (up right) and boreholes (center). 01/2009 to 22/01/2014 was −3.0 °C. Borehole temperature data show warm permafrost (−0.3 °C to Since the installation of the CALM-S, the observational system has −0.9 °C), with thickness varying between 2.5 and 5.0 m (Vieira et al., been continuously improved and the following parameters are current- 2008a; Ramos et al., 2010). The ALT derived from the borehole temper- ly monitored (Tables 1 and 2): ature profiles from 2010 to 2014 varied from 30 to 40 cm. - Air temperatures at 1.60 m above the surface, monitored hourly since 2009. 3. Materials and methods - Ground temperatures in shallow boreholes down to 1.00 m (node 3, 3), 1.60 m (node 7,7) and 4.50 m (node 2,5) (Fig. 3). The boreholes 3.1. Observational setting have a diameter of 32 mm, are cased with PVC pipes, air-filled and temperatures are measured with ibuttons (see Table 2 for specifica- The Crater Lake CALM-S site consists of a 100 × 100 m grid with 121 tions) at different depths starting at 2.5 cm. nodes spaced at 10 m intervals and was installed in January 2006 (Figs. 2 - Ground temperatures in 16 very shallow boreholes regularly distrib- and 3). Thaw depth (TD) is measured manually by mechanical probing uted in the grid, with a single ibutton close to the base of the active (1–2 cm accuracy) once per year in late January or early February (coin- layer (TBAL) at 40 cm depth (Table 2). ciding with the end of summer), depending on logistical constraints - Snow thickness estimated using near surface air temperature since the research station is only open during the summer season miniloggers installed in a stake (5, 10, 20, 40, 80 cm) and a Campbell (Ramos et al., 2007). CC640 time-lapse camera with daily pictures at 11:00, 12:00 and

Fig. 2. Instrumentation setting at the Crater Lake CALM-S site. 522 M. Ramos et al. / Catena 149 (2017) 519–528

Table 1 4. Results “Crater Lake” CALM-S site instrumentation position.

Coordinates Average Snow-stakes: Air Boreholes Time lapse 4.1. Active layer (node 0, 0) altitude snow temperature camera (m a.s.l.) temperature Fig. 4a shows the evolution of thaw depth (TD) at the Crater Lake array and CALM-S grid from 2006 to 2014. The mean TD for the 121 nodes for the mini array whole period was 29.7 cm, but the nine year time series shows a decreasing ′ ″ 62°59 06.7 S 85 2 nodes 1 4.50 m 1 average TD from 35.5 cm in 2006 to 23.1 cm in 2014. Although not constant, 60°40′44.8″ W (3,3) (7,7) node node (2,5) (2,2) 1.00 m with minor increases in TD in 2009 and 2012, the trend is noticeable and node (3,3) followed with higher irregularity by the local maxima and minima mea- 1.60 m sured in individual nodes. Fig. 5 shows the spatial distribution of TD as mea- node (7,7) sured in the grid nodes along the study period. The increasing trend observed in the average values is also clear in the spatial patterns of thaw. Fig. 4b shows the evolution of the air temperatures at the study site from 2009 to 2014. Mean annual air temperature was very stable 13:00 (local solar time). This approach allows evaluating the snow around −3 °C, the same happening with absolute maxima and minima. distribution. The extreme records were +7.8 °C in 2010 and −24.0 °C in 2009. The mean ground temperature close to the base of active layer (TBAL) in the 16 very shallow boreholes from 2008 to 2013 was −1.3 °C, with minor interannual changes and no appreciable trend. The mean maxima were slightly above 0 °C, showing an annual extreme 3.2. Estimation of snow thickness value of 0.5 °C (2008 and 2009) and the minima were always below −4 °C, down to −9.3 °C in 2011. Accounting for the instrumental errors In order to estimate snow thickness, near-surface temperatures ac- of the monitoring devices, the maxima was therefore always close to the quired by an array of sensors installed along a 1.60 m long stake, were water melting point in normal conditions. analyzed. The method implemented by Lewkowicz and Bonnaventure (2008) and based on the changes in thermal regime along the stake 4.2. Permafrost when sensors are inside the snow pack, was used. The daily snow thickness (x ) was classiﬁed according to the follow- i The borehole temperatures from 2010 to 2013 show that the maxi- ing levels: x ≤ 10 cm, 10 cm b x ≤ 20 cm, 20 cm b x ≤ 40 cm, consid- 1 2 3 mum active layer thickness (ALT) calculated using the 0 °C isotherm ering that the periods of snowpack persistence for each level are (t t 1, 2 was 0.40 m with an isothermal transition in the range of 0.4 to 1.2 m, and t ). The Mean Snow Layer Thickness (MSLT) for the entire snow 3 and the mean annual ground surface temperature, at 2,5 cm depth cover period (t = t +t +t ) for each of the snow stakes was calculat- 1 2 3 was −1.7 °C, showing an annual extreme value of 10.7 °C (2013) and ed using the following Eq. (1). the minima were always below −8.1 °C, down to −16.0 °C in 2012. The depth of zero annual amplitude should be slightly below the bottom ½ : þ : þ : of the borehole, just below 4.50 m, with a temperature of ci. −0.9 °C. t1 0 1 t2X0 2 t3 0 4 MSLTðÞ¼ node ðÞm : Fig. 6 shows the thermal envelope for 2010 to 2013 (2012 is not repre- ti ð1Þ i sented due to the data series is not competed).

4.3. Average snow pack

The mean snow pack factor (SF) aims at assessing an annual average The dataset measured by the snow stake installed in the nodes (3,3) and of snow distribution over the CALM-S grid and is an average of the MSLT (7,7), complemented with the pictures taken by the digital camera, allowed at both snow stakes (Eq. (1)). characterizing the evolution of the snow pack from 2008 to 2013 (Fig. 7).

If we account for the number of days of snow pack duration (ti) below the identiﬁed threshold thicknesses (x ≤ 10 cm, MSLTðÞ3; 3 þ MSLTðÞ7; 7 1 SF ¼ ðÞm ð2Þ 10 cm b x2 ≤ 20 cm, 20 cm b x3 ≤ 40 cm), the maximum value of t1 oc- 2 curred in 2010 (137 days) at node (3,3) and the minimum occurred in

Table 2 Crater Lake CALM-S site instrumentation details.

Node Sensor type Accuracy (° Resolution (° Sampling rate Sensor position C) C) (h) (cm)

Air temperature (2,2) Pt-100, 0.2 0.05 1 160 (height) Tinytag Borehole (4.50 m) (2,2) Ibutton 0.5 0.06 3 5, 10, 20, 40, 80, 120, 160, 200, 250, 300, 350, 400, 450 DS1922L (depth) Borehole (3,3) Ibutton 0.5 0.06 3 2.5, 5, 10, 20, 40, 70, 100 (1 m) DS1922L (depth) Borehole (7,7) Ibutton 0.5 0.06 3 2.5, 5, 10, 20, 40, 70, 100, 150 (1.6 m) DS1922L (depth) Temperature close to the base of the 20 m Ibutton 1 0.5 3 or 4 40 active layer interval DS1921G (depth) (16 nodes) Snow stakes (3,3)(7,7) Ibutton 1 0.5 3 5, 10, 20, 40, 80 DS1921G (height) M. Ramos et al. / Catena 149 (2017) 519–528 523

Fig. 4. Thaw depth and temperatures at the Crater Lake CALM-S site from 2006 to 2014. a. Thaw depth, b) Air temperatures, c) Temperature at the bottom of the active layer (40 cm - TBAL).

2008 (66 days) at node (7,7) (Figs. 8a, b). The number of days with corresponds to sectors which are less wind-exposed and with lower re- b20 cm of snow cover (t2) showed a maximum of 175 days in 2013 in lief, showing also a longer and more stable snow cover. This is also node (3,3), while the threshold 40 cm was only attained at node (7,7). shown in the longer lasting and thicker snow pack in node (7,7) than Node (7,7) showed an earlier snow pack setting and later melt, in (3,3) (Figs. 7 and 8). resulting in longer lasting snow pack every year. The snow thickness Despite a slight interannual variability, the decreasing trend in thaw frequently reached over 40 cm at node (7,7), but never in (3,3). In the depth between 2006 (35.5 ± 2 cm) and 2014 (23.1 ± 2 cm) shows a later, snow was generally 10–20 cm thick, but rarely passed 20 cm. In statistically significant coefficient of correlation of 0.85 (Fig. 9). This node (7,7), the snow thickness showed very frequently 20 to 40 cm. trend is consistent with the deceleration in the glacial retreat identified The snow pack generally settled around mid-April to May and lasted in Livingston Island by Navarro et al. (2013) and may be also related to until mid-November to December eve first January, with a significant an increase of snow precipitation pointed by Osmanoglu et al. (2014). delay in the melt date in the more recent years. The impacts of increased snow thickness on a decreasing active layer The results from Eqs. (1) and (2) are shown in Fig. 8c. The larger depth have also been shown by Jiménez et al. (2014). They studied values of SF occurred in 2013 (29 cm) and 2008 (23 cm), and the the thermal behaviour of the active layer by mean of the enthalpy lower in 2011 (17 cm) and 2009 (18 cm). model in function of the snow cover near to the Hurd Peninsula (Living- ston Island). Their results showed a reduction of the soil surface energy 5. Discussion balance when increasing the snow layer thickness. In Byers Peninsula (Livingston Island), De Pablo et al. (2016) studied the evolution of the 5.1. Thaw depth snow cover at the Limnopolar Lake CALM-S site during the 2009-2014 period. Their reported about similar snow depths of about 45 cm, as The analysis of the spatial distribution of thaw depth across the Cra- well as similar snow onset about constant with small variations of ter Lake CALM-S site (Fig. 5) shows that the area with small values 10 days in early March. However, the snow offset had significant 524 M. Ramos et al. / Catena 149 (2017) 519–528

Fig. 5. Summer thaw depth from 2006 to 2014 at the Crater Lake CALM-S site. Scale in decameters.

variations, increasing in the last three years for N60 days. Therefore, sites with different lithological properties under colder climate condi- their main conclusion is that the increase in the snow duration is tions (Hrbáček et al., 2016). resulting in the reduction of the thaw period in the ground. In Byers In our study site in Deception Island the influence of the snow pack Peninsula as well, but in three different monitoring sites with specific on the mean annual thaw depth is well represented by SF through a geomorphological characteristics, Oliva et al. (2016) extrapolated the negative correlation, showing that the increase of snow layer thickness presence of stable frozen layer conditions in the range of 85 to and it persistence period causes a decrease or the maximum thaw 155 cm. These depths agree with the 1.3 m reported for the depth of depth, as shown in Fig. 10. the permafrost table at the nearby Limnopolar Lake CALM-S site by de Pablo et al. (2014). 5.2. Temperature close to the base of the active layer In other localities of the Antarctic Peninsula region, the ALT showed also a high variability, such as on Signy Island, where ALT ranged be- The mean temperature close to the base of the active layer (TBAL) tween 0.8 and 1.8 m influenced by vegetation (Guglielmin et al., was stable during the whole study period (Fig. 4c). During winter and 2012), or on James Ross Island where the ALT reached 0.5 to 0.8 m at spring, the snow cover influences significantly TBAL. Fig. 11 shows M. Ramos et al. / Catena 149 (2017) 519–528 525

Fig. 6. Temperature distribution into the borehole STS-1 (node 2, 5) in the “Crater Lake” CALM-S site. Maximum, minimum and mean annual ground temperatures at different depth are represented (2010, 2011 and 2013, 2012 is not represented because is not complete the data series).

that the years with less days of snow cover (2010, 2011, 2012 and 2014) homogeneously in 2009 (15/04/2009) than in 2011 (18/05/2011), present a good linear fit between the minimum annual air temperature which generated a stronger thermal insulation in the former. This effect and TBAL, implying a poor thermal insulation of the snow pack. On the seems to have contributed to the largest TD in the summer of 2010 other hand, the two years with longer snow pack duration (2009 and (31.5 cm), when compared to the summer of 2012 (28.7 cm). 2013), plot clearly outside the best fit. The annual maxima of TBAL were stable, within 0.0 ± 0.5 °C due to It is also worth comparing 2009 and 2011. The former showed an air the damping effect caused by latent heat fluxes on top of the ice-rich temperature minimum of −24.0 °C and a minimum TBAL of −6.0 °C, permafrost during the thaw season. while the later, showed higher minima (−19.4 °C), but a lower minimum TBAL (−9.3 °C). In this case, SF was very similar in both years 5.3. Permafrost temperatures (18 and 17 cm), but snow cover settled one month earlier and more Maximum annual ground temperatures show similar values and active layer depth of approximately 0.4 m at Crater Lake CALM-S in the 4 years with data. Fig. 6 shows a typical permafrost temperature profile into the borehole STS-1 node (2, 5) at Crater Lake, with maximum summer temperatures nearly isothermal at about 0 °C from 0.4 to 1.2 m depth. This temperature profile reflects the influenceoftheicerichper- mafrost, as well as of the saturated layer just above the permafrost table, which mitigate ground warming. The small interannual range in the maximum mean temperatures for the shallower sensor (2.5 cm depth) at the different years, varying from 8.1 °C in 2010 to 10.7 °C in 2013 is due to the absence of snow during the summer, small interannual air temperature range with similar maxima and latent heat effects due to the high moisture content of the ground. However, the ground temperature minima in this shallow depth (2.5 cm) are very variable inter-annually, with values from −8.1 °C in 2013 to −16.0 °C in 2012, reflecting a strong influence from snow cover variability and differences in winter air temperatures. The mean annual ground temperatures vary very slightly with depth (ci. 1.1 °C). MGST is −1.7 ± 0.5 °C which corresponds to a thermal offset of about 0.8 °C.

5.4. Average snow pack

Snow pack thickness data shows that snow settles between 2 and 4 weeks earlier at node (7,7) than (3,3), which is at a more convex site and closest to the interﬂuve (Figs. 5 and 8). A similar effect occurs Fig. 7. Snow pack thickness at the Crater Lake CALM-S site from 2008 to 2013. a) node at snow melt, with the ground becoming snow free about a week later (3,3), b) node (7,7). at node (7,7). This results in an average of more 27 days per year of 526 M. Ramos et al. / Catena 149 (2017) 519–528

Fig. 8. Characteristics of the snow pack at the Crater Lake CALM-S site from 2008 to 2013. a) Snow pack persistence (ti - days) at node (3,3), b) snow pack persistence in node (7,7), c) Snow pack thickness and Snow Pack Factor (SF).

snow cover difference between the nodes (3,3) and (7,7). MSLT values 6. Conclusions are also higher at (7,7) (25 cm vs 15 cm), reﬂecting the wind sheltered position of that part of the CALM-S site (Fig. 8c). These spatial differ- Ground temperature observations from the Crater Lake CALM-S site ences also affect the spatial distribution of TD (Fig. 5). show that permafrost is warm, with temperatures close to −0.9 °C, with an active layer depth of ci. 40 cm and a thermal offset of 0.8 °C. The zero annual temperature depth is shallow and should be just below 4.5 m depth in contrast to other sites in the Western Antarctic Peninsula in high diffusivity bedrock settings (Ramos and Vieira, 2009; Bockheim et al., 2013; Guglielmin, et al., 2014). These characteristics are strongly inﬂuenced by the high porosity pyroclastic debris of Deception Island, which show a high ice-content at the base of the active layer. Contrary to what would be expected given the general warming signal of the Western Antarctic Peninsula during the last decades, the thaw depth measured during the summer at the Crater Lake CALM-S site showed an approximately linear decreasing trend from 2006 to 2014 at a rate of about 1.5 cm/year. The mean annual air temperatures from 2008 to 2014 were relatively stable at ci −3 °C, therefore showing no clear control on the active layer thinning. The temperatures at 40 cm depth also showed no trend during the same period, which means that the thinning was not due to increase MAAT and also did not result- ed in a cooling at shallow depth. The main factor controlling the decreasing thaw depth was snow cover, with increased duration, especially during the spring and early summer, thus insulating the ground from warming. The reason for this phenomena not having been accompanied by ground cooling is probably related to the high moisture content and ice-rich permafrost in the transient layer, which mitigate temperature change. The results presented in this paper are limited to a small number of years but this is one of the longest data series on thaw depth from an Antarctic CALM-S. The results should be interpreted with caution, but they support the fact that permafrost and active layer dynamics in the Fig. 9. Summer mean thaw depth in Crater Lake CALM-S from 2006 to 2014. Maritime Antarctic is complex and cannot be directly linked to M. Ramos et al. / Catena 149 (2017) 519–528 527

Fig. 10. Maximum thaw depth (TD) as function of the Snow factor (SF) for the Crater Lake CALM-S site between 2008 and 2013. atmospheric warming alone. The changing snow cover plays a major References role in the ground thermal regime and on permafrost and it needs to be better assessed, especially since regional models indicate an increase Anisimov, O.A., Shiklomanov, N.I., Nelson, F.E., 1997. Global warming and active layer in precipitation following atmospheric warming and winter sea ice thickness: results from transient general circulation models. Glob. Planet. Chang. 15, 61–77 (ISSN: 0921-8181). losses in the Western Antarctic Peninsula. Bockheim, J.G., 1995. Permafrost distribution in the Southern Circumpolar Region and its relation to the environment: a review and recommendations for further research. Permafr. Periglac. Process. 6, 27–45. Bockheim, J., Vieira, G., Ramos, M., López-Martínez, J., Serrano, E., Guglielmin, M., Acknowledgements Wilhelm, K., Nieuwendam, A., 2013. Climate warming and permafrost dynamics in the Antarctic Peninsula region. Glob. Planet. Chang. 100, 215–223. http://dx.doi.org/ Authors thank to the Spanish Polar Program, Unit of Marine Technol- 10.1016/j.gloplacha.2012.10.018 (ISSN: 0921-8181). Bojinski, S., Verstraete, M., Peterson, T.C., Richter, C., Simmons, A., Zemp, M., 2014. The ogy from Spanish National Research Council, as well as to the Spanish concept of essential climate variables in support of climate research applications, Army and Spanish Navy crews involved on Antarctic Campaigns for and policy. Am. Meteorol. Soc. (September 2014), 1431–1443. made possible the continuous data acquisition. Our special thanks to Burgess, M.M., Smith, S.L., Brown, J., Romanovsky, V., Hinkel, K., 2000. Global Terrestrial “ ” Network for Permafrost (GTNet-P) permafrost monitoring contributing to global cli- Gabriel de Castilla Spanish Antarctic Station. This research was possi- mate observations. Geol. Surv. Can. 1–7 (Current Research 2000E14). ble thanks to ﬁnancial support of PERMATHERMAL (CONV-2015/001), Brown, J., Hinkel, K.M., Nelson, F.E., 2000. The circumpolar active layer monitoring PERMASNOW (CTM2014-50521-R), ANTARPERMA (CTM2011-15565- (CALM) program: research designs and initial results. Polar Geogr. 24 (3), – E), and PERMAPLANET (CTM2009-10165) research projects from the 165 258. De Pablo, M.A., Ramos, M., Molina, A., 2014. Thermal characterization of the active layer at Spanish National Research Program. the Limnopolar Lake CALM-S site on Byers Peninsula (Livingston Island), Antarctica. Solid Earth 5, 721–739. De Pablo, M.A., Ramos, M., Molina, A., 2016. Snow cover evolution, on 2009-2014, at the Limnopolar Lake CALM-S site on Byers Peninsula, Livingston Island, Antarctica. Cate- na 149, 538–547. Goyanes, G.A., Vieira, G., Caselli, A., Mora, C., Ramos, M., De Pablo, M.A., Neves, M., Santos, F., Bernardo, I., Gilichinsky, D., Abramov, A., Batista, V., Melo, R., Niewendam, A., Ferreira, A., Oliva, M., 2014. Regimen termico y variabilidad espacial de la capa activa en isla decepcion, Antartida. Revista de la Asociación Geológica Argentina. vol. 71 (No 1), 112–124 (ISSN 1851-8249 (en línea). ISSN 1669-7316 (impreso). http:// ppct.caicyt.gov.ar/index.php/raga/issue/view/192/showToc). Guglielmin, M., 2006. Ground surface temperature (GST), active layer, and permafrost monitoring in continental Antarctica. Permafr. Periglac. Process. 17 (2), 133–143. http://dx.doi.org/10.1002/ppp.553. Guglielmin, M., Worland, M.R., Cannone, N., 2012. Spatial and temporal variability of ground surface temperature and active layer thickness at the margin of Maritime Antarctica Signy Island. Geomorphology 155, 20–33. Guglielmin, M., Worland, M.R., Baio, F., Convey, P., 2014. Permafrost and snow monitoring at Rothera Point (Adelaide Island, Maritime Antarctica): implications for rock weathering in cryotic conditions. Geomorphology 225, 47–56. http://dx.doi.org/10. 1016/j.geomorph.2014.03.051. Hrbáček, F., Oliva, M., Laska, K., Ruiz-Fernández, J., de Pablo, M.A., Vieira, G., Ramos, M., Nývlt, D., 2016. Active layer thermal regime in two climatically contrasted sites of the Antarctic Peninsula region. Cuadernos de Investigación Geográﬁca (In press). Jiménez, JJ, Ramos, M, De Pablo, M.A., Molina, A, 2014.Variabilidad térmica de la capa Fig. 11. Minimum temperature close to the base of the active layer (TBAL) versus activa, acoplada al espesor nival, en las proximidades de la BAE Juan Carlos I. Avances minimum air temperature from 2009 to 2014. métodos y técnicas de estudio en el periglaciarismo. Gómez, A; Salvador, F; Oliva, M; 528 M. Ramos et al. / Catena 149 (2017) 519–528

Salvá, M (eds.). Publicacions i edicions de la Universitat de Barcelona. Barcelona temperature sensor (REMS-MSL) on deception island (Antarctica). Cold Reg. Sci. (2014). pp. 287-300. (ISBN-978-84-475-3830-0). Technol. 72, 23–32. http://dx.doi.org/10.1016/j.coldregions.2011.10.012 (ISSN: Lewkowicz, A.G., Bonnaventure, P.P., 2008. Interchangeability of mountain permafrost 0165-232X). probability models, Northwest Canada. Permafr. Periglac. Process. 19, 49–62. http:// Smith, S., Brown, J., 2009. Assessment of the Status of the Development of the Standards dx.doi.org/10.1002/ppp.612. for the Terrestrial Essential Climate Variables - T7 - Permafrost and Seasonally Frozen Marshall, G.J., Lagun, V., Lachlan-Cope, T.A., 2002. Changes in Antarctic Peninsula tropo- Ground. spheric temperatures from 1956 to 1999: a synthesis of observations and reanalysis Styszynska, A., 2004. The origin of coreless winters in the South Shetlands area (Antarcti- data. Int. J. Climatol. 22, 291–310. ca). Pol. Polar Res. 25, 45–66. Meredith, M.P., King, J.C., 2005. Rapid climate change in the ocean west of the Antarctic Smellie, J.L., López-Martínez, J., 2002. Introduction to the geology and geomorphology of Peninsula during the second half of the 20th century. Geophys. Res. Lett. 32, L19604. Deception Island. In: López-Martínez, J., Smellie, J.L., Thomson, J.W., Thomson, Nelson, F.E., Shiklomanov, N.I., 2010. The circumpolar active layer monitoring M.R.A. (Eds.), BAS Geomap Series. British Antarctic Survey, Cambridge, pp. 1–6. network—CALM III (2009–2014): long-term observations on the climate-active Turner, J., Colwell, S.R., Marshall, G.J., Lachlan-Cope, T.A., Carleton, A.M., Jones, P.D., Lagun, layer-permafrost system. In: Blanco, J.J., Ramos, M., de Pablo, M.A. (Eds.), Proceedings V., Reid, P.A., Iagovkina, S., 2005. Antarctic climate change during the last 50 years. Int. of II Iberian Conference of the International Permafrost Association Periglacial, Envi- J. Climatol. 25, 279–294. ronments, Permafrost and Climate Variability. UAH, pp. 9–15 (ISBN: 978-84-9138- Turner, J., Bindschadler, R., Convey, P., et al., 2009. Antarctic Climate Change and the En- 885-5). vironment. Scientiﬁc Committee on Antarctic Research. England, Cambridge. Navarro, F.J., Jonsell, U.Y., Corcuera, M.I., Martín-Español, 2013. Decelerated mass loss of Turner, J., Barrand, N.E., Bracegirdle, T.J., Convey, P., Hodgson, D.A., Jarvis, M., Jenkins, A., Hurd and Johnsons Glaciers. Livingston Island, Antarctic Peninsula. J. Glaciol. 59 Marsall, G., Meredth, M.P., Roscoe, H., Shanklin, J., 2013. Antarctic Climate Change (No. 213). http://dx.doi.org/10.3189/2013JoG12J144. and the Environment: An Update. Polar record. 1-23. http://dx.doi.org/10.1017/ Osmanoglu, B., Navarro, F.J., Hock, R., Braun, M and M.I. Corcuera, 2014. Surface velocity S0032247413000296 ((About DOI), Published online: 18 April 2013). and mass balance of Livingston Island ice cap. Antarctica. Cryosphere, 8, 1807– Vieira, G., Lopez-Martinez, J., Serrano, E., Ramos, M., Gruber, S., Hauck, C., Blanco, J.J., 1823, dói:http://dx.doi.org/10.5194/tc-8-1807-2014. 2008a. Geomorphological observations of permafrost and ground-ice degradation Oliva, M., Hrbacek, F., Ruiz-Fernández, J., de Pablo, M.A., Ramos, M., Vieira, G., Antoniades, on Deception and Livingston Islands, Maritime Antarctica. In: Kane, D., Hinkel, K. D., 2016. Active Layer Dynamics in Three Topographically Distinct Lake Catchments (Eds.), Proceedings of the 9th International Conference on Permafrost, 29 June–3 in Byers Peninsula (Livingston Island, Antarctica). Catena (Accepted). July 2008, Fairbanks, Alaska, Extended Abstracts vol. 1. University of Alaska Press, Ramos, M., Vieira, G., Gruber, S., Blanco, J.J., Hauck, C., Hidalgo, M.A., Tome, D., Neves, M., Fairbanks, pp. 1839–1844 (ISBN 978-0-980017 9-2-2). Trindade, A., 2007. Permafrost and active layer monitoring in the Maritime Antarctic: Vieira, G., Hauck, C., Gruber, S., Blanco, J.J., Ramos, M., 2008b. Massive Ice Detection Using preliminary results from CALM sites on Livingston and Deception Islands. U.S. Geo- Electrical Tomography Resistivity. Examples From Livingston and Deception Islands, logical Survey and The National Academies; USGS OF-2007–1047, Short Research Maritime Antarctic. Proceedings of the VI Assembleia Ibérica de Geofísica, Tomar. Paper 070 http://dx.doi.org/10.3133/of2007-1047srp070. Vieira, G., Bockheim, J., Guglielmin, M., Balks, M., Abramov, A., Boelhouwers, J., Cannone, Ramos, M., Vieira, G., 2009. Evaluation of the ground surface enthalpy balance from bed- N., Ganzert, L., Gilichinsky, D.A., Gotyachkin, S., Lopez-Martinez, J., Meiklejohn, I., rock shallow borehole temperatures (Livingston Island, Maritime Antarctic). Rafﬁ, R., Ramos, M., Schaefer, C., Serrano, E., Simas, F., Sletten, R., Wagner, D., 2010. Cryosphere 3, 133–145 (www.the-cryosphere.net/3/133/2009/). Thermal state of permafrost and active-layer monitoring in the Antarctic: advances Ramos, M., Vieira, G., Guilichinski, D., de Pablo, M.A., Ramos, M., de Pablo, M.A., 2010. during the International Polar Year 2007–2009. Permafr. Periglac. Process. 21 (2), Nuevas estaciones de medida del régimen térmico del permafrost en el área de “Cra- 182–197. http://dx.doi.org/10.1002/ppp.685 (ISSN: 1045-6740). ter Lake”. Isla Decepción (Antártida). Resultados preliminares”. In: Blanco, J.J. (Ed.), World Meteorological Organization, 1997. Global Climate Observing System GCOS/GTOS. Proceedings of II Iberian Conference of the International Permafrost Association Plan for Terrestrial Climate Related Observations. Version 2.0, GCOS-32, WMO/TD, No Periglacial, environments, permafrost and climate variability. UAH, pp. 93–109 796. UNEP/DEIA/TR97-7. WMO, Geneva, Switzerland (130 pages). (ISBN: 978-84-9138-885-5). Ramos, M., De Pablo, M.A., Sebastian, E., Armiens, C., Gómez-Elvira, J., 2012. Temperature gradient distribution in permafrost active layer, using a prototype of the ground